Segmentation of medial temporal lobe subregions in a nearly isotropic space using T2-weighted MRI with anisotropic resolution

Q: Why is T2-weighted MRI segmentation accuracy limited for medial temporal lobe subregions?

High in-plane resolution improves hippocampal contrast, but lower out-of-plane resolution reduces accuracy of subregion thickness measurements.

Q: How is the nearly isotropic segmentation pipeline constructed?

A nearly isotropic atlas is generated by upsampling a high-resolution atlas (from an existing anisotropic atlas) using a nonlocal means approach, and atlas manual segmentations are upsampled using a UNet trained on combined high-resolution ex vivo and low-resolution anisotropic in vivo data.

Q: What improvements are reported for segmentation in isotropic space?

Isotropic T2w segmentation achieved larger effect sizes in the predicted direction and outperformed anisotropic T2w segmentation across most subregions in both cross-sectional group differences and longitudinal consistency analyses.

Sep 30, 2026

Uploaded by Pentious

Background: Morphometry of medial temporal lobe (MTL) subregions on brain MRI serves as a sensitive imaging biomarker for Alzheimer’s Disease and related conditions. Using T2-weighted (T2w) MRI with high in-plane resolution supports hippocampal subfield segmentation, but limited out-of-plane resolution degrades subregion thickness measurement accuracy. Objective: Develop and validate a nearly isotropic segmentation pipeline that improves thickness accuracy in T2w MRI by upsampling images and labels and applying high-resolution segmentation.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages4

Reading Time10 min

AccessFree to download

Abstract

Background

Method

Result

Conclusion

DOI: 10. 1002/alz70856_105240 BIOMARKERS POSTER PRESENTATION NEUROIMAGING Segmentation of medial temporal lobe subregions in a nearly isotropic space using T2-weighted MRI with anisotropic resolution Yue Li1 Pulkit Khandelwal 1 Long Xie2 Nidhi S. Mundada 1 Christopher A Brown 1 Laura E. M. Wisse3 Emily McGrew 1 Amanda E Denning 1 Sandhitsu R. Das 1 DavidA. Wolk 1 Paul A. Yushkevich 1 1 University of Pennsylvania, Philadelphia, PA, USA 2Siemens Healthineers, Princeton, NJ, USA 3 Lund University, Lund, Sweden Correspondence Yue Li, University of Pennsylvania, Philadelphia, PA, USA. [Email:](Email: yue.li@pennmedicine.upenn.edu)[ yue.li@pennmedicine.upenn.edu](Email: yue.li@pennmedicine.upenn.edu) Abstract Background: Morphometry of medial temporal lobe (MTL) subregions in brain MRI is sensitive biomarker to Alzheimer’s Disease and other related conditions. While T2-weighted (T2w) MRI with high in-plane resolution is widely used to segment hippocampal subfields due to its higher contrast in hippocampus, its lower out-of-plane resolution reduces the accuracy of subregion thickness measurements. To address this issue, we developed a nearly isotropic segmentation pipeline that incorporates image and label upsampling and high-resolution segmentation in T2w MRI. Method: First, a high-resolution atlas was created based on an existing anisotropic atlas derived from 29 individuals. Both T1-weighted and T2w images in the atlas were upsampled from their original resolution to a nearly isotropic resolution using a nonlocal means approach. Manual segmentations within the atlas were also upsampled to match this resolution using a UNet-based neural network, which was trained on a cohort consisting of both high-resolution ex vivo and low-resolution anisotropic in vivo MRI with manual segmentations (Figure 1a). Second, a multi-modality deep learningbased segmentation model was trained within this nearly isotropic atlas (Figure 1b). This method was evaluated on independent sets, including cross-sectional (N = 196) and longitudinal (N = 31) MRI scans, which were used for the group difference analysis (Amyloid+ mild cognitive impairment(A+ MCI)vs. Amyloid-cognitively normal(A-CN)) and longitudinal consistency analysis, respectively (Figure 1c). Result: Table 1(a) displays the group differences of cross-sectional median thickness between A+MCI and A-CN with age as covariate. The T2w segmentation in isotropic space achieved larger effect sizes in the predicted direction (A+MCI < A-CN) and outperformed T2w anisotropic segmentation over most subregions. Table 1(b) This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2025 The Alzheimer’s Association. Alzheimer’s & Dementia published by Wiley Periodicals LLC on behalf of Alzheimer’s Association. Alzheimer’s Dement. 2025;21(Suppl. 2):e105240. [https://doi.org/10.1002/alz70856_105240](https://doi.org/10.1002/alz70856_105240) wi[leyonlinelibrary.com/journal/alz](leyonlinelibrary.com/journal/alz) 1of4 BIOMARKERS shows the consistency analysis of longitudinal median thickness. When measured as the sum of absolute median thickness differences, the consistency of isotropic T2w segmentation outperformed that of anisotropic T2w segmentation over most subregions. Figure 2 shows the visualization of the segmentation and point-wise group difference analysis at different resolutions, with isotropic T2w segmentation demonstrating a smoother surface and larger effect sizes than anisotropic T2w segmentation. Conclusion: Nearly isotropic subregion segmentation improved the accuracy of cortical thickness as an imaging biomarker for neurodegeneration inT2w MRI. BIOMARKERS BIOMARKERS

FAQs

Why is T2-weighted MRI segmentation accuracy limited for medial temporal lobe subregions?

High in-plane resolution improves hippocampal contrast, but lower out-of-plane resolution reduces accuracy of subregion thickness measurements.

How is the nearly isotropic segmentation pipeline constructed?

A nearly isotropic atlas is generated by upsampling a high-resolution atlas (from an existing anisotropic atlas) using a nonlocal means approach, and atlas manual segmentations are upsampled using a UNet trained on combined high-resolution ex vivo and low-resolution anisotropic in vivo data.

What improvements are reported for segmentation in isotropic space?

Isotropic T2w segmentation achieved larger effect sizes in the predicted direction and outperformed anisotropic T2w segmentation across most subregions in both cross-sectional group differences and longitudinal consistency analyses.

Sep 30, 2026

Uploaded by Pentious